Intrusion detection in healthcare domain using machine learning

Abstract

Intrusion detection system is considered as one of the main sources for protection of important information and communication technologies especially in healthcare networks. Intrusion detection system must be updated frequently because of intrusions that improves periodically. Due to particular limitations such as resource-constrained devices, limited memory and battery capacity of nodes, and unique protocol stacks, traditional intrusion detection algorithms must be updated and improved for application to the Internet of Things. To address this issue, a lightweight attack detection approach based on supervised machine learning–based FIDS (Frothy Disturbance Intrusion Detection System) was developed to detect an adversary attempting to inject unneeded data into the network. FIDS can play an important role in intrusion detection. The proposed model was trained with KDD (Knowledge Discovery in Database) dataset using SVM (Support Vector Machine) algorithm. AODV (Ad-hoc- on-demand Distance Vector) routing protocol was used for routing to make it more energy efficient since the energy usable was limited. The process of packet transmission was done with the help of clustering algorithm to separate the sensors into groups with each group containing a cluster head.